Steven Cole, Robert Moore, Alice Robson,
Paul Mattingley and Vicky Bell
BHS Meeting: Hydrology for Flood Risk Management 21 November 2013
Recent developments in
countrywide flood forecasting
using the G2G distributed
●
Summer 2007 floods
£3 billion insurance payouts 55,000 properties flooded,
36,000 from surface water flooding
National infrastructure impacts
140,000 homes without clean water for 17 days 42,000 homes without power for 24 hours
●
Pitt Review commissioned
Flood Forecasting Centre & Scottish Flood Forecasting Service Countrywide flood forecasting using G2G distributed model
Lumped catchment model
Distributed Hydrological Modelling (G2G)
―
Catchment-wide storm―
Lower catchment storm―
Upper catchment stormGrid-to-Grid (G2G) Model
• One model for large regions (UK) • Flow estimates in each grid (1km2)
• Realistic response to different storm intensities and extent
Gauging station Benefits Flo w m 3 s -1 G2G Time (days) 30% increase!
Soil properties (for each HOST class)
Terrain/slope
Steep slopes → Shallow storage
Shallow slopes → Deep storage
Soil Moisture Deficit (modelled) Gridded rainfall G2G Runoff Production Urban/Suburban coverage • Increase runoff (by
decreasing soil depth)
Raingauge-adjusted radar rainfall (HyradK)
G2G national application for FFC & SFFS
Price et al.; Cranston and Tavendale, Proc. ICE – Water Management (2012)
River flow Soil moisture
G2G runs 24/7 within FEWS across England, Wales & Scotland
Operates at a 15 minute time-step on a 1km grid, out to 5 days
Forecasts river flow, surface runoff and soil moisture
Flood Guidance Statement (FGS)
•G2G forecasts inform Flood Guidance Statement
•Daily flood risk assessment for emergency responders (10:30) •Updated during events
•Information presented at county level out to 5 days
m3s-1
Normal 2 year flood
More extreme
1km maps
G2G flow outputs converted to flood return periods using
G2G for Scotland
Cranston and Tavendale, Proc. ICE – Water Management (2012)
•Recent improvement in local forecasting model coverage
•BUT still large areas of Scotland
without a flood forecasting capability
•Flood Risk Management (Scotland) Act 2009
•Created joint Met Office and SEPA
Scottish Flood Forecasting Service
•G2G applied to give first national
picture of flood risk
G2G for Scotland – Snowmelt modelling
Model Performance
(R2)
G2G for Scotland – Snowmelt modelling
No Snowmelt With Snowmelt Model Performance (R2) Major improvement in model performanceG2G for Rapid Response Catchments
Area <100km2
All stations
R2 Efficiency
• Explore use of G2G for Rapid Response Catchments
• typically small area (<100km2) & ungauged• need to develop forecast/warning capability
• Value of radar/NWP rainfall forecast ensembles? • Case study experience & methods of display
G2G simulation (Jan – Apr 2008)
G2G for Rapid Response Catchments
• Analysis of G2G forecasts using:
• Raingauge data as foreknowledge of forecast rainfall data • UKV deterministic NWP as forecast rainfall data
• Suggests spatial biases over 2010/11 assessment period
August 2010 – July 2011 Large circles denote NWP
based forecast worse than
• Post-processing scheme to generate large rainfall ensembles cheaply • Seamless combination of nowcast and deterministic NWP forecast • Noise used to generate ensembles and downscale NWP
• 12 members (now 24), 15 minute accumulations,
7h Nowcast every 15 mins, 24h Blended ensemble every 6h
Forecast time origin
Q(T) return period threshold used
Circles denote gauging stations Solid outline: area <50km2
Observed flow exceeds threshold during forecast
Percentage of ensembles that exceeded the Q(T) threshold at some point during forecast
0 0-33% 34-66% >67%
Case study: 6-7 July 2012
●
Slow moving depression over
south-west England
●
>100mm recorded in 18hr
ending 12:00 7 July 2012
●
River Axe
severely affected, and south Cornish coast
Threshold
Forecast
Origin
Q
med/2 Q
medQ(5) Q(50)
Case study: 6-7 July 2012
06-07-2012 07:15
Threshold
Forecast
Origin
Q
med/2 Q
medQ(5) Q(50)
Case study: 6-7 July 2012
06-07-2012 07:15 06-07-2012 19:15 07-07-2012 00:15 07-07-2012 07:15
Threshold
Forecast
Origin
Q
med/2 Q
medQ(5) Q(50)
Case study: 6-7 July 2012
06-07-2012 07:15 06-07-2012 19:15 07-07-2012 00:15 07-07-2012 07:15 Early signal of flood risk “hotspots”
Case study: 6-7 July 2012
Yealm at Puslinch (55km2) Axe at Chard Junction (85km2)
07-07-2012 00:15
07-07-2012 07:15
●
Strong signal at least 12 hours before
06-07-2012 19:15 06-07-2012 07:15
●
Extreme Rainfall Alerts (ERA)
● National rainfall-threshold based method ● Based on FEH 30 year return period
rainfalls “averaged” across 8 UK cities
●
G2G runoff production
affected by:
● Rainfall amount plus
● Urban/suburban coverage ● Soil and geology properties
● Antecedent soil moisture conditions
●
Prototype runoff threshold
exceedances seem more targeted
1h radar rainfall totals >30mm >25mm 1h runoff totals >8.5mm >7mm
G2G alerts for surface water flooding
Real-time SWF impact modelling approach
Hazard
Footprint
Based on G2G surface runoffHazard
Impact
Maps
Exposure
e.g. Property, Population, Infrastructure National Receptor Dataset (EA) National Population Database (HSL) Updated FloodMap for Surface Water (EA)
Natural Hazards Partnership (NHP)
1500 1600 1500
Population
G2G Hazard footprint
1600 1700 1700SWF Population impacts
Vulnerable Pop.
Key High Hazard Hazard Key High Hazard Population Count 4,900 (MAX) 1 Key Hazard Population Count 18,000 (MAX) 1 1500 1600 1700 High Hazard Vulnerable Pop Count1,150 (MAX) 1
Key
Hazard
Vulnerable Pop count 3,000 (MAX) 1
Hazard impact
maps
Closing remarks
●
Step change in flood forecasting capability across Britain
G2G provides fluvial forecasts “everywhere” on a 1km grid Probabilistic forecasts over several days possible
●
G2G has shown utility for Rapid Response Catchments
Uses high-res deterministic and probabilistic rainfall products Case studies shows potential for early warning (possibly 12h+)●
Real-time mapping of surface water flooding impacts
Aim to move from static to dynamic maps of hazard and impact Presentation of results, local, regional and national (with HSL)
